Hector Sanjuan
Hector Sanjuan
@lanzafame what's the status of this? If we decide not to go this way we should close the issue.
The `allocations` option and the replication factor allow to control where things should be pinned, and by default peers with less space have the least priority. If you don't want...
One of the issues here is that Cluster does not know (or cares) about the size of the DAG it is pinning. It would be possible to parse Unixfs root...
> Isn't there a way to ask ipfs what the actual size of a CID is? There is no way. The only way to know how much space a CID...
This is not a fix, but related: https://github.com/ipfs/ipfs-cluster/pull/1656
hmm thanks, it seems we are going to have to start automatically increasing the fd limit like IPFS. Alternatively the badger defaults we use in the configuration may need tweaking...
Maybe related to: ``` // Reduce this from 64MiB to 16MiB. That means badger will hold on to // 20MiB by default instead of 80MiB. DefaultBadgerOptions.MaxTableSize = 16
@vasa-develop feel free to unassign if you dropped this
We fixed the issue with peers re-joining after network problems. I think most points here are addressed by `--consensus crdt` and the new rpc authorization. Definitely 1,3,5 and 6.